{
  "path": "/tutorials/langchain-tokenba-api",
  "reviewedAt": "2026-10-23",
  "version": "langchain-openai1.7.2/langchain-core1.7.0",
  "status": "partial-local-native-pass-gateway429",
  "sourceURLs": [
    "https://docs.langchain.com/oss/python/integrations/chat/openai",
    "https://docs.ollama.com/api/openai-compatibility"
  ],
  "validationScope": "真实LangChain LCEL经本地模型解析JSON7/5/12、退出0；随后同一真实SDK使用apizt HTTPS/gemini-2.5-flash、max_tokens64/max_retries0/timeout30单次请求明确HTTP429当前分组上游饱和，停止且未重投。不能把本地通过当公网网关通过。",
  "bodySteps": [
    "在虚拟环境安装langchain-openai，固定完整依赖版本。",
    "ChatOpenAI显式设置本地base_url=http://127.0.0.1:11435/v1、model=tutorial-qwen3-proof、api_key=ollama（本地忽略的占位值）。",
    "把7支蓝笔与5支红笔的原创输入交给PromptTemplate，用LCEL三段链invoke。",
    "保存真实模型原始字符串、SDK版本与退出码，再独立比较total=12；格式不合规需按失败展示。",
    "切换apizt时必须另行使用其真实Base URL/Key/准确模型ID，不能用本地通过证明公网账户和模型已通过。",
    "公网分支显式ChatOpenAI base_url=https://api.apizt.com/v1，api_key只读环境变量，model=gemini-2.5-flash，max_tokens64、max_retries0、timeout30。本次真实返回429当前分组上游饱和；保存错误requestID并停止，输出token上限不是费用上限。"
  ],
  "files": [
    {
      "file": "README.md",
      "bytes": 779,
      "sha256": "ea51cede882d00905d869b799c33075085f4b3957a05ee46a1d6099a933baa02"
    },
    {
      "file": "input.json",
      "bytes": 95,
      "sha256": "8f0add780ec1363b398bcdfed6f2c39758247d8c6fdd2d3cd8c9dfe49755fdbe"
    },
    {
      "file": "tokenba-execution.json",
      "bytes": 4938,
      "sha256": "71a9016f6b9dfa303cb982c420b6b61231d7cb47d0ea3fc074a1d4dc389fdfc4"
    },
    {
      "file": "requirements.txt",
      "bytes": 845,
      "sha256": "1c58c544638674053ee72e82af2ebde1b3b06bb0fcf8a02eee5ffcee682988ac"
    },
    {
      "file": "run-execution.json",
      "bytes": 353,
      "sha256": "5e8179f9978194c9bca5ad5141bb4e1724938cbae9340b4ae745e5136fb31407"
    },
    {
      "file": "run-tokenba.py",
      "bytes": 713,
      "sha256": "769515dc987c84f5effec45c6e773761e935278159c192e16db467ceeff3c34a"
    },
    {
      "file": "run.py",
      "bytes": 838,
      "sha256": "d97a0d13eaf76a944cc48839e2019b310468a09a65dd15f1ce19e0964ff9c0af"
    }
  ],
  "privateCredentialsPublished": false,
  "commands": [
    "python -m pip install -r requirements.txt",
    "python run.py"
  ],
  "observations": [
    "本地真实LCEL链返回并解析blue7/red5/total12，exit0。",
    "真实apizt HTTPS/gemini-2.5-flash、max_retries0请求明确HTTP429当前分组上游饱和。",
    "公网路径已停止，本地通过不等于公网通过。"
  ],
  "limits": [
    "只说明当前测试Key/分组/模型这一时点；不是全站不可用诊断。",
    "输出token上限不是费用上限；未验证公网成功生成或工具能力。"
  ],
  "tokenbaTestKeyUsed": true,
  "timeoutInferenceUsage": "本地路线；没有使用apizt测试Key。",
  "modelProvenance": "隔离Ollama0.40.9；模型为官方Qwen3:0.6b完整GGUF，经原生ollama create导入为tutorial-qwen3-proof，非完整OCI pull成功。权重SHA256 7f4030143c1c477224c5434f8272c662a8b042079a0a584f0a27a1684fe2e1fa，522640096字节；原生自动识别模板，capabilities tools/thinking/completion。",
  "gatewayAttemptScope": "仅当前测试Key的Gemini分组；model gemini-2.5-flash；BaseURL HTTPS api.apizt.com。仅认证到该HTTPS域名，未改全局配置，无重投。不能扩成全站/所有分组/所有模型不可用。"
}